T cell receptors (TCRs) are key mediators of the immune response to disease, with significant therapeutic potential. For example, clinical trials have shown that adoptive transfer of T cells genetically engineered to express TCRs of defined antigen specificity can lead to tumor regression in vivo. However, the therapeutic capabilities of TCRs are limited due to their relatively weak binding affinities, and the efficacy of genetically engineered T cells is limited by the mispairing of transduced TCR chains with those endogenously produced. Efforts to improve TCR affinity via in vitro molecular evolution can impact antigen specificity, and attempts to improve chain pairing have yet to yield an effective solution. Here we propose to develop a computational modeling framework to engineer TCRs with high affinity, high antigen specificity, and biased chain pairing. Structure-guided computational design will be used to comprehensively scan interface mutations, model their impact on specificity and affinity, and tailor designs to target key antigenic determinants. This work will be accomplished through the close interaction of two laboratories with complementary expertise: protein interaction modeling and structure-based design of TCRs (PI Zhiping Weng) and experimental TCR biophysics (PI Brian Baker). Experimental assessments, including measurements of affinities and immunological assays of function and specificity, will be used to iteratively test and improve the modeling framework. The framework and associated algorithms will be publically released, facilitated via a web-accessible database serving as a repository for measurements of wild type and mutant TCR binding affinities and structures. The proposal has three specific aims: Aim 1. Design point mutations in TCRs to improve affinity and specificity towards antigen; Aim 2. Perform flexible redesign of TCR loops with multiple substitutions to further improve and redirect binding; Aim 3. Bias specific pairing of engineered TCR chains via computationally designed constant regions.